Head Pose Normalization for Recognition of Human Identities Using Color and Depth Data

When using today's admission control systems some kind of interaction is generally required. The idea of this work was to approach identification in a contact-less manner and to improve the aspect of comfort at a relatively low loss of accuracy by recognizing people from visual facial features while they walk towards the entrance. Ideally people will not have to interact with the system deliberately. Two well known holistic methods for facial recognition are Principal Component Analysis (PCA) or Eigenfaces [17] and Local Binary Pattern Histograms [4]. Such holistic methods are expected to profit from precise alignment of face images. In this work a method will be presented to generate precisely aligned facial color and depth images. The recognition results show that both holistic methods profit significantly from these normalization steps.